How often do trains run on time?
84.4% of passenger trains in Great Britain arrived on time over the 13 four-week periods to August 2026. The average passenger arrived 3.4 minutes late, and 5.4% of planned trains were cancelled in the latest four-week period.
Moving annual average to August 2026. down 0.7 points on July 2026.
Minutes late, average passenger, year to August 2026.
Of planned trains, latest four weeks to August 2026.
Punctuality over time
What’s this?
The share of passenger trains in Great Britain that arrived at their final stop on time, defined as within 5 minutes (10 for long-distance services), over the latest 13 four-week periods.
What it cannot tell you: A cancelled train counts as late, but a train that is on time at the end can still have been late at stations on the way. It counts trains, not passengers, so a busy train counts the same as an empty one.
What’s this?
The average number of minutes late that a passenger arrives, counting trains that run to time as zero, over the latest 13 four-week periods.
What it cannot tell you: It is an average across all passengers, so a few very late journeys are hidden by many on-time ones.
What’s this?
The share of planned passenger trains in Great Britain that were cancelled in each four-week period, counting a part-cancelled train as a fraction.
What it cannot tell you: Strikes, storms and engineering work push up cancellations in some periods. It counts trains, not passengers affected.
Period by period
Great Britain, moving annual averages.
| Period | Trains on time (public performance measure) | Average passenger lateness |
|---|---|---|
| August 2026 | 84.4% | 3.4 |
| July 2026 | 84.8% | 3.3 |
| June 2026 | 85.0% | 3.3 |
| May 2026 | 85.2% | 3.2 |
| April 2026 | 85.5% | 3.1 |
| March 2026 | 85.5% | 3.1 |
| February 2026 | 85.6% | 3.1 |
| January 2026 | 85.6% | 3.1 |
| December 2025 | 85.5% | 3.1 |
| November 2025 | 85.4% | 3.1 |
| October 2025 | 85.2% | 3.2 |
| September 2025 | 85.2% | 3.2 |
Questions people ask
What counts as on time?
A train is on time if it arrives at its final stop within 5 minutes of the planned time (10 minutes for long-distance trains). A cancelled train counts as not on time. This is called the public performance measure.
Does this include Scotland and Wales?
It covers all of Great Britain: England, Scotland and Wales. Northern Ireland's railways are separate.
Why use a moving annual average?
Single four-week periods swing with weather, strikes and engineering work. The annual average over the latest 13 periods smooths those swings so the trend is easier to see.
What does “trains on time (public performance measure)” measure?
The share of passenger trains in Great Britain that arrived at their final stop on time, defined as within 5 minutes (10 for long-distance services), over the latest 13 four-week periods. It was 84.4% (% of trains, moving annual average) in August 2026, down 7.6 points on January 2021. The highest we have is 93.4% (March 2021) and the lowest 84.4% (August 2026).
What does “average passenger lateness” measure?
The average number of minutes late that a passenger arrives, counting trains that run to time as zero, over the latest 13 four-week periods. It was 3.4 (minutes late, moving annual average) in August 2026, up 94% on January 2021. The highest we have is 3.4 (August 2026) and the lowest 1.4 (March 2021).
What does “trains cancelled” measure?
The share of planned passenger trains in Great Britain that were cancelled in each four-week period, counting a part-cancelled train as a fraction. It was 5.4% (% of planned trains (four-week periods)) in August 2026, up 2.8 points on January 2021. The highest we have is 8.0% (December 2022) and the lowest 0.9% (April 2020).
Sources: Office of Rail and Road, passenger rail performance. Updated every night. Each chart has a CSV and a Share button.
See also: Train and bus fares · Road deaths